ti mmwave in building automation · 2019. 12. 5. · tuning of parameters in ti design enables...
TRANSCRIPT
TI Information – Selective Disclosure TI Information – Selective Disclosure
TI mmWave in Building Automation
For 2019/11/13 Seminar
End Equipment Overview IP Network Camera
Motion Detectors
Automated Doors & Gates
People Counting
1
TI Information – Selective Disclosure
mmWave Sensors – Technology Overview
2
• Motion Detection – Indoor detection of breathing, talking, typing for lighting control and security. Outdoor detection at extended distances (50m+) for security. Motion classification of people vs animals vs background.
• People Counting – Counting and location tracking of multiple people for security, retail, elevators, factories.
• Automated Doors – Tracking of objects approaching automatic door. Classification of object direction, size.
• IPNC Surveillance – Augmenting security cameras for enhanced motion detection or point-tilt-zoom (PTZ) control
When to use
mmWave
sensing?
• mmWave technology is robust against environmental influences such as dazzling sunlight, no light, weather conditions like rain and fog and changes in temperature
• Detect very fine motions, RF technology for penetration through materials like plastic, fabric, and drywall. Use in privacy conscious applications
• RFCMOS technology enables analog/digital integration in a single low-power, small, single-chip solution
Why Now?
• mmWave is the band of spectrum between 30GHz and 300GHz
• Electromagnetic waves used for sensing, imaging and communications
• mmWave sensors measure with high accuracy range, velocity and angle of remote objects
What is
mmWave
sensing
TI Information – Selective Disclosure
1. A synthesizer (synth) generates a “chirp"
2. The chirp is transmitted by the TX antenna
3. The chirp is reflected off an object and the reflected chirp is received
at the RX antenna.
4. The RX signal and TX signal are ‘mixed’ and the resulting signal is called an
‘IF signal’.
1 2
3
4
IF signal
Synth
TX ant.
RX ant. mixer
Tx = sin 𝑤1𝑡 + ϕ1) IF = sin 𝑤1 − 𝑤2 𝑡 + (ϕ1−ϕ2)
Rx = sin 𝑤2𝑡 + ϕ2)
A
t fc=77GHz
t Tc=40us
f 81GHz
B=4GHz
S
Basics of FMCW (Frequency Modulation Continue Wave)
The fundamentals of millimeter wave sensors, http://www.ti.com/lit/wp/spyy005/spyy005.pdf
3
TI Information – Selective Disclosure
Basics of FMCW (Range Measurement)
By working with FFT on these IF signals to get Range, Velocity, Angle
information of detecting object
IF frequency = Tx frequency – Rx frequency
IF frequency
4
TI Information – Selective Disclosure
Basics of FMCW (Velocity and Angle Measurement)
Velocity and Angle of object reflects in phase difference of IF signal.
IF frequency = Tx frequency – Rx frequency
Multiple
Transmission chirps
separated in time
Multiple chirps for velocity detection
Reception of chirps over different
antennas separated in space
(distance)
Multiple received chirps.
Reflected Signal from
moving object has
different phase for two
reflected chirps.
(Intermediate frequency)
Multiple antennas for angle detection
5
TI Information – Selective Disclosure
TI Innovation – Single-Chip CMOS
6
Discrete Multi-Chip mmWave Sensor • Discrete solution – expensive
• Complex and critical signal routes
• Unconventional packaging
• Prone to noise
• Lack of system level monitoring
• Crude implementation of RF and Baseband safety
TI Single-Chip mmWave Sensor • Smaller in size
• Simpler design
• Built in monitoring and calibration (SIL)
• High Resolution, less false positives
• Programmable core
• Lower Power
PCB1 PCB2 PCB
mmWave
RX+TX+ADC
MCU + DSP
TI Information – Selective Disclosure
IWR1xxx mmWave Signal Processing
7
RF Front-End ADC ADC Data
IWR1443
IWR6843
Pre-
Processing (Interference Mitigation)
1st Dim FFT
(Range)
2nd Dim FFT
(Velocity)
3rd Dim FFT
(Angle Arrival) Detection
Point Cloud [Range, Velocity, Angle]
Clustering Tracking Object Classification Objects
TI Information – Selective Disclosure
Industrial Radar – Path of Development
8
Discover
“I am looking for a new sensor for my
project”
Evaluate
“I have purchased an EVM to evaluate
mmWave sensors”
Design
“I have selected mmWave sensors for my
project”
• Tools & SW that provide building
blocks to develop unique mmWave
sensor applications
• Enable rapid development, test and
deployment
mmWave SDK & LIB
IWR1xxxBOOST EVM
App Notes, TID & Training
• Tools & SW focused on shortening
evaluation time by simplifying getting
started and providing reference
software to start with
mmWave Labs
Chirp Database
OOB Demo + Visualizer
mmWave Studio
• Tools focused on showcasing
mmWave sensor capabilities and
potential
• Enable Proof of Concept validation of
use cases
Sensing Estimator
TI.com Portal
White Papers & Videos
mmWave Experiments
mmWave Sensors Forum (e2e)
TI Information – Selective Disclosure
Object Range Detection
9
TI Confidential – NDA Restrictions Traffic monitoring Intelligent street lighting
Level sensing
Forklifts Robotics
Drones
People counting
Perimeter
security
IWR mmWave Sensors TI's single chip mmWave sensors integrate
a DSP, MCU and RF front-end to detect
range, velocity and angle
Measure tank fluid level
with unprecedented
accuracy for inventory
control and early leak
detection
Sensing performance that improves pedestrian safety and provides
power/cost savings through intelligent triggering of lighting
Detect objects in
obstructed views for
intelligent safety
Unprecedented
accuracy at the
micrometer level
Enable autonomous flight for
building, land surveying and
delivering packages
Detect people in a zone of
interest and trigger actions
Enabling security
systems with motion
sensitive detection
and tracking
Detect traffic location and
volume more accurately
Doors & Gates Intelligent perimeter
detection
TI Information – Selective Disclosure
Building Automation – Problems to Solve using mmWave
11
False Detection – Fine Motion
• Sensing systems to detect occupancy today are prone to false detection
• In security, all detection events must be responded to by a person (camera, guard, police)
• Costs $$$, false detection desensitize responders to real threats
• Need sensing systems that can ignore common false detection sources (environment, objects, movement outside ROI)
People Localization and Counting
• Locating and counting people can be done today, but requires complex and expensive technologies like stereo vision and 3D ToF
• Doing this accurately requires expensive processing and complex SW.
• Need for less expensive, higher accuracy solutions
• TI mmWave on-board processing enables counting and tracking of multiple people in single-chip
Motion Classification
• Classification of objects and motion such as determining if John, a forklift, or a dog has just moved into a ROI
• Can be done today, but requires complex and expensive technologies like vision processors.
• Need for less expensive solutions that require less complex SW implementations
• TI mmWave range, velocity, angle, data along with point cloud size can be used for classification
TI Information – Selective Disclosure
mmWave Sensors for Building Automation – Offering Summary
12
Application Occupancy Detection and
Automated Doors and Gates
Motion Detectors and IP Network
Cameras
Benefits Track and separate multiple people
simultaneously with a single chip sensor. Simplify
solution by ignoring non-moving (static) objects
like chairs and tables and filtering moving objects
such as fans, curtains, or blinds.
Robust detection and localization of slow
moving people even in cluttered outdoor
areas with lots of movement from clutter like
trees and shrubs
Hardware
(field of view)
IWR6843 ISK EVM
(120° horizontal, 30°
vertical)
IWR6843 ODS EVM
(120° horizontal,
120° vertical)
IWR6843 ISK EVM
(120° horizontal, 30° vertical)
Example
Environments
Conf
Room
Open
Office
Open Office
(ceiling mount)
Outdoors
Example Coverage
Area
6m x 6m 14m x
14m
16m x 16m 25m x 50m
Example Range
(resolution)
6m
(0.048m)
14m
(0.12m)
8m
(0.12m)
55m
(0.5m)
Example Velocity
(resolution)
5.17m/s
(0.082m/s)
5.25m/s
(0.082m/s
5.25m/s
(0.082m/s)
6.2m/s
(0.1m/s)
Applicable
Onboard Algorithms
Static Clutter Removal,
Group Tracking/Counting
Static Clutter Removal,
Group Tracking for Object filtering
Provided Example
Demonstrations
• People Counting
Lab
• Indoor False
Detection Mitigation
Lab
• Overhead People
Counting Lab
50m Outdoor False Detection Filtering Lab
Discover
ti.com and
videos
1. Review mmWave Building Automation
Applications page
2. Watch the People Counting Video
3. Watch the People Counting Applications Video
4. Watch the Intelligence at the Edge Video
Evaluate
Evaluation
Kits
1. Order IWR6843ISK + MMWAVEICBOOST EVM
2. Order IWR6843 ODS Antenna Board
3. Order mmWave POE Board
Experiments
1. Fine Motion Detection vs PIR Experiment
2. People Counting through drywall, glass, wood
Experiment
3. Response of radar to rain Experiment
4. *NEW* Fall Detection with IWR6843
Labs
1. People Counting Lab
2. Indoor False Detection Mitigation Lab
3. 50m Outdoor False Detection Filtering Lab
4. Overhead People Counting Lab
Reference
Designs
1. People Counting and Tracking Reference Design
using IWR6843
Design Explore
Device
1. Product Folder IWR6843
2. Reference IWR6843 datasheet, errata and TRM
3. Review IWR6843 EVM schematics and layout
Follow the links below for where to go on
ti.com
TI Information – Selective Disclosure
L: Conference Room with Static Clutter Removal for chairs and table
R: Hallway Scene person in GREEN tracked at 14m with Medium Range
Configuration and Group Tracking
13
Base configurations of people counting TI Design support 6m and
14m operation.
Tuning of parameters in TI Design enables variety of applications
and environments
1. Discover mmWave offering for people tracking and counting page here
1. Watch Video: People Counting Applications & Benefits
2. Watch Video: Intelligence at the Edge
2. Evaluate the performance
1. Order IWR6843 EVM here
2. Download People Counting Lab
3. Download Indoor False Detection Mitigation Lab
3. Design custom boards with IWR6843 silicon
1. Reference IWR6843 datasheet, errata and TRM
2. Review IWR6843 EVM schematics and layout
4. Leverage turn-key or design custom solutions using 3rd Parties such as:
1. Turn-Key and Custom: Ainstein
2. Turn-Key and Custom: D3 Engineering
3. Turn-Key and Custom: Smart Radar Systems (SRS)
Mounting assumes 1.5-2.5m elevation, with 10 degree downtilt
Wall Mounted People Tracking and Counting Reference Design using mmWave Radar Sensor TIDEP-01000, Design Status: On ti.com
Short Range
Configuration
Medium Range
Configuration
HW / EVM IWR6483 ISK EVM
Field of View 120° Horizontal, 30° Vertical
Max Range 6m 14m
Example Area 6m x 6m 6m x 14m
14m x 14m
Range Resolution 4.8cm 12cm
Max Velocity 5.17 m/s 5.25 m/s
Velocity Resolution 0.082 m/s 0.082 m/s
Algorithms Used Static Clutter Removal, Group
Tracking, False Detection
Mitigation
Static Clutter Removal, Group
Tracking, False Detection
Mitigation
System Power ~1.5W
Location accuracy Person location within <16cm
Counting density 3 persons per square meter
Demonstrated accuracy +/- 0 persons +/- 1 persons +/- 2 persons
3 people in scene >95% of frames 100% of frames 100% of frames
5 people in scene >51% of frames >85% of frames 100% of frames
7 people in scene >59% of frames >85% of frames >98% of frames
9 people in scene >14% of frames >43% of frames >84% of frames
TI Information – Selective Disclosure 14
Ability to detect height of people and classify as
standing/sitting/laying down (YELLOW – standing, BLUE – sitting)
Mounting and sensing distance assumes 3m elevation
POE enables simplified integration with existing infrastructure
March 2019
(Release on TI REX)
HW / EVM IWR6843 ODS EVM
IWR6843 Power Over Ethernet Adaptor
Field of View 120° Horizontal, 120° Vertical
Max Range 4m – radial
Example Area 6m x 6m
Range Resolution 12cm
Max Velocity 5.25m/s
Velocity Resolution 0.082m/s
Algorithms Used Static Clutter Removal, Group Tracking, Stance Detection
System Power ~1.5W
Performance Details Max count: 3 people
Stances detected: standing / sitting
Base configurations of ceiling mounted people counting TI Design
support 360° radial operation.
Tuning of parameters in TI Design enables variety of applications
and environments
Ceiling Mounted People Tracking and Counting Reference Design using mmWave Radar Sensor and POE Design Status: Available on TI REX
Electrical Power
Data + Configuration
IEEE
802.3a
f Class
1 POE
1. Discover mmWave offering for people tracking and counting page here
1. CES 2019 : Video Standing / Sitting detection
2. CES 2019 : Video Conference Room Implementation with 3P Ainstein
2. Evaluate the performance
1. Order IWR6843 ODS EVM + MMWAVEICBOOST
2. Order mmWave POE Board
3. Download Overhead People Counting Lab
3. Design custom boards with IWR6843 silicon
1. Reference IWR6843 datasheet, errata and TRM
2. Review IWR6843 EVM schematics and layout
4. Leverage turn-key or design custom solutions using 3rd Parties such as:
1. Custom: Ainstein
2. Turn-Key and Custom: RF Beam
TI Information – Selective Disclosure
Intelligent Motion Detection
• Only turn on camera if radar detects and verifies motion
• Reduce false detection, less false alarms
• Result is system resource conservation:
– Reduce Power Consumption
– Reduce Network Bandwidth – more cameras in system
– Reduce Video Storage – less server storage required
15
Radar ON: Low Power
State
Motion Detected
and classified by Radar
Video Stream
Turned ON
Vision Fusion / PTZ Control
• Use of both camera vision and radar combined to determine position and
velocity of people
• Use radar to identify targets even in rain, fog, dust, and other extreme
conditions
• Locate and track targets for PTZ and focus control
Outdoor 50m People Tracking and False Detection Mitigation Application Usage
TI Information – Selective Disclosure
Windy day testing showing false detection filtering with Group
Tracking algorithm
16
Example Configuration
HW / EVM IWR6843 ISK EVM (ES2.0)
Max Field of View /
Coverage Area
120° Horizontal, 30° Vertical
Max Range 56m
Range Resolution 50cm
Max Velocity 6.2m/s
Velocity Resolution 0.10m/s
Algorithms Used Static Clutter Removal,
Group Tracking (configured for outdoor, false detection filtering)
System Power ~1.5W (not duty cycled)
100mW-500mW (duty cycled, 1 frame to 5 frames)
Performance Metrics • Max range: 56 m departing, 56 m approaching
• Sneaking person (Crouched) detected @56m departing and
@42m approaching
Field of View (horizontal)
at measured distance
55m +/- 45°
40m +/- 60°
30m +/- 60°
20m +/- 60°
10m +/- 60°
Detects slow moving people even in cluttered outdoor areas with
lots of motion
• Robust sensing regardless of environment, uses integrated DSP on
IWR6843 and algorithms to ignore point cloud from dynamic clutter
like trees and shrubs
• Use with IP Network Cameras to turn cameras on or send alerts
when a moving person is detected only Moving Trees/Shrubs
seen but filtered
X-Y Scatterplot View
Cumulative
Range vs Doppler View
See person starting
at ~55m
Person Tracked
@ 50m
Outdoor 50m People Tracking and False Detection Mitigation
1. Discover mmWave offering for people tracking and counting page here
1. Watch Video: Intelligence at the Edge
2. Evaluate the performance
1. Order IWR6843 ISK EVM here
2. Download 50m People Tracking and False Detection Mitigation Lab
3. Design custom boards with IWR6843 silicon
1. Reference IWR6843 datasheet, errata and TRM
2. Review IWR6843 EVM schematics and layout
4. Leverage turn-key or design custom solutions using 3rd Parties such as
1. ODM: Alpha Networks
2. ODM: Primax
3. Custom: Smart Radar Systems (SRS)
4. Custom: Colorado Engineering (CEI)
TI Information – Selective Disclosure
Evaluate today!
Detecting Human Falls and Stance with IWR6843
17
Shows the change in shape and height of mmWave 3D point cloud when person
stands, sits, or lays down
• Uses wide-angle antenna on IWR6843 ODS EVM to visualize the point cloud of
person standing, sitting, or laying down
• Point cloud and velocity information could be used to create algorithm to determine if
person has fallen down for applications such as elderly monitoring
• Visualization of stance was demonstrated at ranges of 2m and 5m
More information:
including
Documentation
and Software
Available on TI
Resource
Explorer
Hardware IWR6843 ODS
EVM
Base Truth
Color
indicates
height
relative to
EVM
Standing Sitting Laying Down
EVM
(backside)
3D point
cloud
TI Confidential – NDA Restrictions
Building Automation – Technology Comparison
18
Passive Infrared Cameras
Pros:
• Simple, low power consumption (uA-mA)
Cons:
• Low sensitivity to motion
• False detection outdoors from sunlight, temperature
• Limited range (5-10m), no position/range information
TI mmWave Radar
Pros:
• Algorithms applied for variety of applications
• Video for recording and monitoring
Cons:
• False detection from shadows, occlusion, day/night cycles,
changing environment.
• No position/range information
• Privacy considerations
Measures change in infrared light to detect
motion
Video image processor analyzes imagery to
determine people movement and behavior
Active Infrared (3D ToF, LIDAR)
Pros:
• High sensitivity to motion (breathing, typing)
• Simple static and dynamic object separation
• Onboard DSP processing for single-chip tracking,
classification of objects
• Extended range for person detection (50m+)
• Insensitive to weather, changing environments
Cons:
• Lower angular resolution than camera or active infrared
TI’s fully-integrated, single-chip 77GHz and
60GHz mmWave radar
Pros:
• High angular resolution provides rich dataset similar to
camera
Cons:
• Limited range in presence of sunlight (5-10m)
• Requires substantial processing to separate and classify
relevant objects
• System complexity (optics, illumination, processing)
Measurement of infrared light time of flight
TI Information – Selective Disclosure
IWR6843 Industrial Starter Kit
19
Antenna
Board
Carrier
Board
Antenna
Parameters
IWR6843 ISK IWR6843 ODS IWR6843 AOP
Azimuth FOV +/- 60 degree +/- 60 degree +/- 65 degree
Elevation FOV +/- 20 degree +/- 60 degree +/- 65 degree
Azimuth Angle
Resolution 15 degree 29 degree 29 degree
Elevation Angle
Resolution 58 degree 29 degree 29 degree
Max Expected
Distance
(Human)
50m 20m 25m
Bandwidth 3.5GHz
(60.5 – 64GHz)
1.7GHz
(61 – 62.7GHz)
4GHz
(60 – 64GHz)
• Carrier board with swappable antenna
boards
• Swappable antenna boards support
different antenna designs
TI Information – Selective Disclosure
DCA1000 Raw Data Capture Solution
20
• Enables RF performance evaluation with PC environment
• mmWave ADC data can be captured and brought onto PC for advanced algorithm
development and prototyping
• Data capture (recording) via Ethernet interface at high speed ( ADC, User data,
debug data)
• mmWave Studio software tool used to visualize object range/velocity/angle
x W R 1 x + D C A 1 0 0 0
or